Medical errors have been prominently discussed in the national media in the last decade. The landmark study by Brennen et al1 in 1991 showed that adverse events occurred in 3.7% of hospitalizations and that 27.6% of these adverse events were due to medical error. In obstetrics, particularly, there were adverse events in 1.5% of hospitalizations; 38.3% of these adverse events were due to medical error. Childbirth is the second most common reason for hospitalization, with nearly 4 million deliveries in the United States annually.2 Thus, we estimate that there are approximately 22,980 adverse events caused by medical error in obstetric hospitalizations each year.
Although many factors have been implicated as causes for medical error, one of the most commonly cited reasons is fatigue of medical providers. There is increasing evidence that sleepiness adversely impacts clinical performance.3 Despite this evidence, with the notable exception of the State of New York, there has been very little regulation of physician work hours.
In response to the concerns that restricted sleep may have had a detrimental effect on patient safety, the Accreditation Council of Graduate Medical Education (ACGME) placed limits on the shift length and total hours worked by house staff, effective July 1, 2003.4 These regulations have brought into place sweeping schedule changes in many training programs. Work-hour reforms compliant with ACGME regulations were instituted at MetroHealth Medical Center in July of 2002. These new regulations offer the opportunity for a natural experiment regarding the effect of house staff hours on the quality of obstetric care provided. Therefore the specific aim of our study is to measure the effect of house staff work hours on the quality of obstetric and gynecologic care.
Our study seeks to compare sentinel events, medication errors, patient outcomes, decision making, and patient perceptions between the 2 shift-type eras.
MATERIALS AND METHODS
In the Department of Obstetrics and Gynecology at MetroHealth Medical Center, there have been 2 distinct eras of house staff shift schedules. Before July 2002, house staff call lasted 36 hours, on an average of every fourth night. There was a second-year night float system in place Sunday through Thursday, and the other on-call residents slept an estimated 1–2 hours per on-call night. There were no restrictions on the number of hours worked per week. Effective July 2002, all residents were scheduled, on average, to work no more than 24 hours per shift and no more than 80 hours per week. There was a night-float team (consisting of a postgraduate year 2 resident and a senior resident) Sunday through Thursday. The interns took 24-hour call on average every fourth night and went home in the morning post-call. Therefore, we defined “before” ACGME work-hour reforms as July 2001 to June 2002, and we defined “after” ACGME work-hour reforms as July 2002 to June 2003. There was 24-hour in-house attending coverage both before and after work-hour reforms. The seniority level of residents on call remained unchanged before and after the work-hour reforms.
After obtaining a Health Insurance Portability and Accountability Act of 1996 waiver and Institutional Review Board approval, we obtained data from the MetroHealth perinatal database, medication error database, incident reports filed in the legal department, and the quality management database.
The perinatal database includes every patient who delivers a baby at MetroHealth Medical Center. It includes information on patient demographics and outcomes for mother and baby. We examined patient age, parity, race, insurance status, gestational age, delivering and attending physician, and infant weight to estimate that the population has not changed significantly over time. We also estimated the rates of third- and fourth-degree perineal lacerations, fevers in labor, postpartum hemorrhage, umbilical arterial pH less than 7.0, need for general anesthesia at cesarean delivery, primary cesarean delivery rate, “code pink” resuscitations, admissions to the neonatal intensive care unit (NICU), and Apgar scores less than 7 at 5 minutes. To examine decision making on the part of the physician, we compared indications given for cesarean delivery for trends before and after work-hour reforms were implemented.
The medication error database is a hospital pharmacy–maintained database of all medication errors that are detected. The identified cases are classified as errors pertaining to prescribing, administering, transcription, or dispensing. We limited our investigation to prescribing errors. Because the manner in which the database was collected changed in January of 2002, we limited the medication error database to January through April 2002 and January through April 2003 to ensure that we were studying comparable data sets.
Incident reports are expected to be filed in the legal department/office of risk management whenever a medical or hospital system error is recognized. Two reviewers (J.L.B. and M.H.B.) independently examined all incident reports in the hospital from July 2001 to June 2003. Reviewers determined whether the incident reports involved an obstetrics–gynecology resident and whether the resident was an identifiable contributor to the incident. Discrepancies between reviewers were resolved by mutual agreement. Reviewers were not blinded to year of occurrence.
After any hospitalization at MetroHealth Medical Center, a subset of patients is randomly selected to participate in telephone surveys conducted by an independent firm (Quality Data Management, Cleveland, OH) to assess consumer satisfaction. These consumer comments make up the Quality Desktop database. Two reviewers independently assessed the consumer comments from obstetric and gynecologic patients. Discrepancies were resolved by mutual agreement.
All comparisons were made with Student t tests and Pearson χ2 where appropriate.
Patients were similar in age, race, number of pregnancies, and gestational age at delivery between the 2 time periods (Table 1). There was a significant drop in the number of patients considered to be staff patients. This was likely due to the hiring of faculty whose private practice consisted of patients formerly considered staff patients. Because there were no changes in the patient demographics other than their private versus staff designation, there were no changes in patient demographic characteristics.
There was a significant decrease in the proportion of neonatal resuscitations and maternal hemorrhage after work-hour reform (Table 2). With the exception of increased cesarean delivery for chorioamnionitis, none of the other maternal or neonatal outcomes we examined were significantly different (Table 3). Policy change regarding intubation for meconium in the neonatology division over the 2 years may help explain the decrease in neonatal resuscitations. Increases in primary cesarean delivery rates approached significance. However, this finding is consistent with the national increases in primary cesarean delivery rates and may not be related to resident work-hour reforms, because most babies in the United States are not born in teaching hospitals.5
There was a single report of a prescribing error. Although this likely represents underreporting, we were unable to compare prescribing errors before and after work-hour changes. Similarly, incident reports filed in the legal department involving residents were rare (13 over 2 years). There were 3 the first year and 10 the second year. There were too few incidents to determine statistical differences between the 2 years. National rates for medical error in an obstetrics admission are 0.5%. With the volume of deliveries at our institution, we would expect 17 errors per year. Thirteen incident reports over 2 years suggests that errors are underreported.
The Quality Desktop database showed no trends in complaints about residents over the 2 years studied. Most comments from patients focused on structural elements of the hospital or clinic, eg, parking limitations or nursing staff.
Although several maternal and neonatal outcomes improved with the decrease in resident working hours, overall there was not an impressive difference in quality of care before and after work-hour reform. The sensitivity of the databases we used to detect differences in medical error varied, and we did not have the power to detect differences in incident reports, prescribing errors, or patient satisfaction. Although differences in maternal and neonatal outcomes were detected, incident reports and prescribing errors were likely underrepresented.
We examined the effect of resident physicians’ work-hour changes on quality of care and not private attending work-hour changes. Residents, by design, have more quality checks on their performance and, therefore, more opportunity for mistakes to be caught before patient harm is done. Thus, our study findings that quality of care was not improved with work-hour reform should not be generalized to suggest that private physician work-hour restrictions would have the same results.
There is abundant evidence that fatigue contributes to poorer job performance in areas outside of medicine.6 Both acute sleeplessness and chronic sleeplessness are a problem.3 Sleepiness is thought to have contributed to some of the major disasters of our time, such as the Chernobyl and Three Mile Island disasters.7 In professions in which sleepiness is believed to affect performance, such as long-distance trucking and the airline industry, there are regulations in place to limit the amount of time worked and the number of shifts worked per week.
Although the data on clinical performance in study settings clearly shows worsening performance with sleep deprivation, other papers have questioned the improvements in patient care with multiple shift changes.3 Cross-coverage of patients by physicians unfamiliar with the patient may lead to more errors.8 Thus, although physicians may be more rested, the errors associated with the increase in shift changes that occur with shorter shifts may offset the gains to be made from decreasing sleep deprivation. The net effect of resident work-hour reform on patient outcomes and medical error is unknown.
Quality of obstetric care may be defined in many different ways. Quality of care is more than just the absence of errors. Quality of care may be less than ideal and still not drop below a given threshold level. Because it is not clear exactly which tasks in obstetrics–gynecology would be most affected by fatigue, we chose 5 major categories to measure quality: sentinel events, medication errors, patient outcomes, decision making, and patient perceptions. By looking at a variety of measures of quality, we hoped to better ascertain how fatigue affects house staff performance. However, incident reports, prescribing error reporting, and consumer feedback databases proved to have low sensitivity and inadequate power for detecting these errors. Repeating our study with medical record review or physician self-reporting of errors might improve the detection of medical errors.9
Although problems in physician performance may be underreported, implementing resident work-hour restrictions shows minimal evidence of improvement in quality of care.
1. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, et al. Incidence of adverse events and negligence in hospitalized patients. N Engl J Med 1991;324:370–6.
2. Popovic J, Hall M. 1999 National Hospital Discharge Survey. Advance data from vital and health statistics; no. 319. Hyattsville (MD): National Center for Health Statistics; 2001.
3. Weinger M, Ancoli-Isreal S. Sleep deprivation and clinical performance. JAMA 2002;287:955–7.
4. ACGME Work Group on Resident Duty Hours. Report of the ACGME work group on resident duty hours. Chicago (IL): Accreditation Council for Graduate Medical Education; 2002.
5. Phibbs C, Bronstein J, Buxton E, Phibbs R. The effect of patient volume and level of care at the hospital of birth and neonatal mortality. JAMA 1996;276:1054–9.
6. Jha A, Duncan B, Bates D. Fatigue, sleepiness, and medical error. In: Shojania K, Duncan B, McDonald K, editors. Making health care safer: a critical analysis of patient safety practices. Evidence Report/Technology Assessment No. 43. (Prepared by the University of California at San Francisco-Stanford Evidence-Based Practice Center Under Contract No. 290-97-0013), AHRQ Publication No. 01-E058. Rockville (MD): Department of Health and Human Services; 2001.
7. Mitler M, Carskadon M, Czeisler C, Dement W, Dinges D, Graeber R. Catastrophes, sleep, and public policy: consensus report. Sleep 1988;11:100–9.
8. Petersen LA, Brennan TA, O'Neil AC, Cook EF, Lee TH. Does housestaff discontinuity of care increase the risk for preventable adverse events? Ann Intern Med 1994;121:866–72.
9. O'Neil AC, Petersen LA, Cook EF, Bates DW, Lee TH, Brennan TA. Physician reporting compared with medical-record review to identify adverse medical events. Ann Intern Med 1993;119:370–6.